Microwave blood glucose sensor data processing method, device, equipment, medium and product

Abstract

The invention discloses a microwave blood glucose sensor data processing method, device and equipment, a medium and a product, and relates to the field of microwave noninvasive blood glucose monitoring, the method comprises the following steps: obtaining sampling data of a microwave blood glucose sensor; converting the sampling data into an original image; according to the original image, a resolution improvement model is adopted to obtain a resolution optimized image; the resolution improving model is a super-resolution generative adversarial network pre-established according to a training sample set; and converting the resolution optimization image into resolution optimization data. According to the method, the original sampling data of the microwave blood glucose sensor are converted into the original image, the original image is converted into the resolution optimization image by building the super-resolution generative adversarial network, the resolution optimization data are further obtained, the problem that the quality factor of the sensor is reduced due to insufficient sampling points is solved, and the quality factor of the sensor is improved. Resonance frequency point errors caused by insufficient sampling points are reduced, and the data resolution of the microwave blood glucose sensor is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of microwave non-invasive blood glucose monitoring, in particular to a microwave blood glucose sensor data processing method, device, equipment, medium and product. BACKGROUND

[0002] Microwave non-invasive blood glucose monitoring is a non-invasive method that measures blood glucose concentration by utilizing the dielectric properties of the interaction between microwave signals and blood. Compared with traditional blood glucose monitoring methods (such as finger blood sampling or implantable blood glucose monitoring devices), microwave non-invasive blood glucose monitoring technology has the advantages of comfort, portability, no risk of infection, low cost, etc., and is convenient for continuous non-invasive blood glucose monitoring.

[0003] However, microwave non-invasive blood glucose monitoring technology also faces some challenges. Microwave sensors based on resonance aim to measure blood glucose concentration by analyzing the changes in resonance frequency points, but in actual application, due to the large size and high price of vector network analyzers, in order to realize the wearability of the system, it is usually impossible to integrate all the functions of the vector network analyzer into the circuit. This leads to the problem of insufficient sampling points due to the limitation of power consumption and other conditions of the integrated circuit, which further leads to a significant decrease in the quality factor of the sensor, making it difficult to accurately distinguish small frequency differences, thereby affecting the resolution of the sensor. SUMMARY

[0004] The purpose of the present application is to provide a microwave blood glucose sensor data processing method, device, equipment, medium and product, to solve the problem of decrease in sensor quality factor caused by insufficient sampling points, to reduce the resonance frequency point error caused by insufficient sampling points, and to improve the data resolution of the microwave blood glucose sensor.

[0005] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a microwave blood glucose sensor data processing method, comprising:

[0007] obtaining sampling data of a microwave blood glucose sensor;

[0008] converting the sampling data into an original image;

[0009] According to the original image, a resolution optimization image is obtained by using a resolution enhancement model; the resolution enhancement model is a super-resolution generative adversarial network built in advance according to a training sample set; the training sample set includes multiple original sample images and a resolution optimization sample image corresponding to each original sample image;

[0010] convert the resolution optimization image into resolution optimization data.

[0011] Optionally, the sampling data is converted into an original image, specifically comprising:

[0012] According to the sampling data, an index mapping method is adopted to obtain a two-dimensional index matrix;

[0013] The two-dimensional index matrix is subjected to data normalization processing to obtain a scaling matrix; the data range of the scaling matrix is between 0 and 255;

[0014] According to the scaling matrix, a color mapping algorithm is adopted to obtain the original image.

[0015] Optionally, the method for building the resolution enhancement model comprises:

[0016] A plurality of blood glucose samples of different concentrations are prepared, and microwave signal original data corresponding to each blood glucose sample is obtained respectively;

[0017] For each microwave signal original data, low sampling rate and low resolution data and high sampling rate and high resolution data of the microwave blood glucose sensor are obtained respectively; the low sampling rate and low resolution data is a sampling result with a number of sampling points less than 100, and the high sampling rate and high resolution data is a sampling result with a number of sampling points greater than 200;

[0018] Each low sampling rate and low resolution data is converted into an original sample image, and each high sampling rate and high resolution data is converted into a resolution-optimized sample image;

[0019] Each original sample image is subjected to sample augmentation to obtain an input image set, and each resolution-optimized sample image is subjected to sample augmentation to obtain an output image set;

[0020] According to the input image set and the output image set, a super-resolution generative adversarial network is trained to obtain the resolution enhancement model.

[0021] Optionally, before obtaining, for each microwave signal original data, low sampling rate and low resolution data and high sampling rate and high resolution data of the microwave blood glucose sensor, the method for building the resolution enhancement model further comprises:

[0022] The microwave signal original data is sequentially subjected to abnormal point removal and smoothing filtering processing.

[0023] Optionally, each original sample image is subjected to sample augmentation to obtain an input image set, and each resolution-optimized sample image is subjected to sample augmentation to obtain an output image set, specifically comprising:

[0024] Each original sample image is subjected to image flipping, contrast adjustment, saturation adjustment, and Gaussian noise addition processing to obtain the input image set;

[0025] The output image set is obtained by performing image flipping, adjusting contrast, adjusting saturation, and adding Gaussian noise processing on each resolution-optimized sample image respectively.

[0026] Optionally, each residual block of the super-resolution generative adversarial network comprises, in sequence, a convolution layer, a batch normalization layer, a Mish activation function, a convolution layer, a batch normalization layer, and a self-attention module.

[0027] In a second aspect, the present application provides a microwave blood glucose sensor data processing device, applied to the microwave blood glucose sensor data processing method described above, comprising:

[0028] A sampling module is configured to acquire sampling data of a microwave blood glucose sensor.

[0029] An image conversion module is configured to convert the sampling data into an original image.

[0030] A resolution improvement module is configured to obtain a resolution-optimized image according to the original image by using a resolution improvement model; the resolution improvement model is a super-resolution generative adversarial network pre-built according to a training sample set; the training sample set comprises multiple original sample images and resolution-optimized sample images corresponding to each original sample image.

[0031] A data conversion module is configured to convert the resolution-optimized image into resolution-optimized data.

[0032] In a third aspect, the present application provides a computer device, comprising a memory, a processor, a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the microwave blood glucose sensor data processing method described above.

[0033] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the microwave blood glucose sensor data processing method described above.

[0034] In a fifth aspect, the present application provides a computer program product comprising a computer program, and the computer program is executed by a processor to implement the microwave blood glucose sensor data processing method described above.

[0035] According to the specific embodiments provided by the present application, the present application has the following technical effects:

[0036] The application provides a microwave blood glucose sensor data processing method, device, equipment, medium and product, converts sampling data of a microwave blood glucose sensor into an original image, converts the original image into high-resolution data after improving the resolution of the original image, improves the data resolution by improving the image resolution through mutual conversion between data and images; the low sampling rate and low-resolution image is converted into a low sampling rate and high-resolution image by building a super-resolution generative adversarial network, and high-resolution data is further obtained, the problem of sensor quality factor decline caused by insufficient sampling points is solved, the resonance frequency point error caused by insufficient sampling points is reduced, and the data resolution of the microwave blood glucose sensor is improved. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Figure 1 An application environment diagram of a microwave blood glucose sensor data processing method in an embodiment of the present application;

[0039] Figure 2 A flowchart of a microwave blood glucose sensor data processing method provided by an embodiment of the present application;

[0040] Figure 3 A two-dimensional index matrix diagram corresponding to some sampling data provided by an embodiment of the present application;

[0041] Figure 4 A three-channel image diagram corresponding to the two-dimensional index matrix; Figure 3 A three-channel image diagram corresponding to the two-dimensional index matrix;

[0042] Figure 5 A flowchart of a resolution improvement model building method provided by an embodiment of the present application;

[0043] Figure 6 A flowchart of a resolution improvement model building method provided by an embodiment of the present application; Figure 4 An image diagram obtained after image flipping processing;

[0044] Figure 7 An image diagram obtained after image flipping processing; Figure 4 An image diagram obtained after adjusting the contrast;

[0045] Figure 8 An image diagram obtained after adjusting the contrast; Figure 4 An image diagram obtained after adjusting the saturation;

[0046] Figure 9 An image diagram obtained after adjusting the saturation; Figure 4A schematic diagram of the image obtained after adding Gaussian noise;

[0047] Figure 10 This is a schematic diagram of the structure of a super-resolution generative adversarial network provided in an embodiment of this application;

[0048] Figure 11 This is a schematic diagram of a high-resolution sample image provided in an embodiment of this application;

[0049] Figure 12 To and Figure 11 A schematic diagram of the resulting image obtained by interpolating and enlarging the corresponding low-resolution original sample image using the Bicubic algorithm;

[0050] Figure 13 To and Figure 11 A schematic diagram of the resulting image obtained by processing the corresponding low-resolution original sample image through the super-resolution generative adversarial network of this application;

[0051] Figure 14 A functional module diagram of a microwave blood glucose sensor data processing device provided in an embodiment of this application;

[0052] Figure 15 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] This application proposes a method, apparatus, device, medium, and product for processing microwave blood glucose sensor data. It converts the raw sampling data of the microwave blood glucose sensor into raw images, and then converts the raw images into high-resolution images with low sampling rates by building a super-resolution generative adversarial network, and further obtains high-resolution data. This can solve the problem of sensor quality factor degradation caused by insufficient sampling points, reduce resonant frequency errors caused by insufficient sampling points, and improve the data resolution of microwave blood glucose sensor.

[0055] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] The microwave blood glucose sensor data processing method provided in this application embodiment can be applied to, for example... Figure 1The application environment shown. Among them, the microwave blood glucose sensor is connected with the ESP32 single-chip microcomputer, the sampling data of the microwave blood glucose sensor is obtained by setting the corresponding sampling frequency, and the single-chip microcomputer converts the sampling data into an original image, that is, an original low sampling rate and low resolution image. According to the original image, a super-resolution generative adversarial network is adopted according to the training sample set to obtain a resolution optimized image, that is, a low sampling rate and high resolution image, and finally the low sampling rate and high resolution image is converted into a high resolution data.

[0057] In one exemplary embodiment, as Figure 2 shown, a microwave blood glucose sensor data processing method is provided, comprising steps 201 to 204. Among them:

[0058] Step 201, obtaining the sampling data of the microwave blood glucose sensor. In this embodiment, the sampling data of the microwave blood glucose sensor S 21 parameters are obtained by the single-chip microcomputer integrated system. S 21 Parameters are the forward transmission coefficients of the microwave network, which are used to describe the transmission of the microwave signal from the input end to the output end.

[0059] Step 202, converting the sampling data into an original image.

[0060] Step 203, according to the original image, a resolution enhancement model is adopted to obtain a resolution optimized image. The resolution enhancement model is a super-resolution generative adversarial network built in advance according to the training sample set. The training sample set includes multiple original sample images and resolution optimized sample images corresponding to each original sample image.

[0061] Step 204, converting the resolution optimized image into resolution optimized data. According to the resolution optimized data, more accurate blood glucose monitoring results can be obtained, and the data resolution of the microwave blood glucose sensor is improved.

[0062] In one exemplary embodiment, Figure 2 Step 202 in the above embodiment can be replaced by the following steps:

[0063] Step 2021, according to the sampling data, an index mapping method is adopted to obtain a two-dimensional index matrix.

[0064] Step 2022, the two-dimensional index matrix is subjected to data normalization processing to obtain a scaling matrix. The data range of the scaling matrix is between 0 and 255.

[0065] Step 2023, according to the scaling matrix, a color mapping algorithm is adopted to obtain an original image.

[0066] In this embodiment, the microwave frequency corresponding to the sampling data and S 21The values of the parameters constitute a two-dimensional matrix, and a two-dimensional index matrix as shown in Figure 3 is obtained through an index mapping method. The index mapping method can map the microwave frequency and S 21 parameters of the sampling data to the corresponding positions of the two-dimensional index matrix, and the spatial distribution information of the sampling data can be preserved to the maximum extent through index mapping.

[0067] In order to adapt to the input requirements of the architecture of the super-resolution generative adversarial network, it is necessary to convert the two-dimensional matrix into a three-channel image (RGB image). To this end, first, the two-dimensional index matrix is subjected to data normalization processing to obtain a scaling matrix, so that the data range is between 0 and 255. Then, the 'jet' color mapping algorithm is used to convert the scaling matrix into an RGB image and save it as a PNG file. Figure 3 The two-dimensional index matrix in Figure 4 is converted to obtain a three-channel image as shown. The converted three-channel image will be used as the input of the super-resolution generative adversarial network for super-resolution reconstruction.

[0068] In an exemplary embodiment, as shown in Figure 5 , the method for building the resolution enhancement model includes the following steps 501 to 505. Among them:

[0069] Step 501, prepare a plurality of blood glucose samples of different concentrations, and obtain the corresponding microwave signal original data of each blood glucose sample. The blood glucose samples of different concentrations can be blood samples or glucose solution samples.

[0070] Step 502, for each microwave signal original data, obtain low sampling rate low resolution data and high sampling rate high resolution data of the microwave blood glucose sensor, respectively. Among them, the low sampling rate low resolution data is a sampling result with less than 100 sampling points, and the high sampling rate high resolution data is a sampling result with more than 200 sampling points.

[0071] Step 503, convert each low sampling rate low resolution data into an original sample image, and convert each high sampling rate high resolution data into a resolution optimized sample image.

[0072] Step 504, sample expansion is performed on each original sample image to obtain an input image set, and sample expansion is performed on each resolution optimized sample image to obtain an output image set.

[0073] Step 505, according to the input image set and the output image set, the super-resolution generative adversarial network is trained to obtain a resolution enhancement model.

[0074] In an exemplary embodiment, Figure 5Before step 502 in the method, the method further comprises: sequentially performing abnormal point removal and smoothing filtering processing on the microwave signal raw data.

[0075] In an exemplary embodiment, glucose solutions with different concentrations are prepared as blood glucose samples. A calibrated electronic balance is used to weigh glucose with concentrations of 0 mg / dL-500 mg / dL (concentration increment of 10 mg / dL) in sequence, and mix them with 500 mL of distilled water in a beaker. In order to reduce the influence of solution inhomogeneity, an electric stirrer, a liquid circulating pump and a polyvinyl chloride (PVC) plastic tube are used to form a complete liquid circulating system to keep the glucose solution in the tube fully uniform.

[0076] The plastic tubes containing glucose solutions with different concentrations are fixed above the blood glucose sensor by adhesive tape, and a voltage stabilizer is used to power the active part of the blood glucose sensor. The ESP32 single-chip microcomputer integrated system is used to measure the S 21 values. The Digital-to-Analog Converter (DAC) module of the ESP32 single-chip microcomputer provides the control voltage of the Voltage-Controlled Oscillator (VCO), and the radio frequency signal generated by the VCO is transmitted to the envelope detector after passing through the microwave blood glucose sensor. Finally, the output voltage of the envelope detector is sampled by the Analog-to-Digital Converter (ADC) module of the ESP32 single-chip microcomputer and transmitted to the host computer through the single-chip microcomputer. The above-mentioned glucose solutions have 51 concentrations, and 5 data acquisitions are performed at each concentration, resulting in 255 groups of microwave signal raw data. The microwave sensor is used to collect the S 21 values of the microwave signal raw data corresponding to different concentrations of glucose solution at multiple discrete frequency points. For each microwave signal raw data, by modifying the single-chip microcomputer program, the S 21 values at 36 discrete frequency points and 288 discrete frequency points are sequentially collected as low sampling rate low resolution data and high sampling rate high resolution data, respectively. In this way, 255 groups of low sampling rate low resolution data and 255 groups of high sampling rate high resolution data corresponding to each other are obtained.

[0077] In order to enhance the diversity of the data set and improve the robustness of the model, the sample set is expanded. In an exemplary embodiment, Figure 5 Step 504 in the method can be replaced by the following steps:

[0078] Step 5041, image flipping, adjusting contrast, adjusting saturation and adding Gaussian noise are performed on each original sample image respectively to obtain an input image set.

[0079] Step 5042, image flipping, adjusting contrast, adjusting saturation and adding Gaussian noise are performed on each resolution-optimized sample image respectively to obtain an output image set.

[0080] Thus, four additional sample images are obtained for each sample image. Taking the three-channel image in Figure 4 as an example, Figure 4 the processing results after image flipping, adjusting contrast, adjusting saturation and adding Gaussian noise are shown in Figures 6 to 9 respectively. After sample set expansion, the final input image set is composed of 1275 original sample images of 36x36 pixels; the final output image set is composed of 1275 resolution-optimized sample images of 288x288 pixels. The input image set and the corresponding output image set are divided into a training set, a test set and a validation set according to a ratio of 8:1:1.

[0081] In an exemplary embodiment, a super-resolution generative adversarial network containing a self-attention mechanism is constructed. The super-resolution generative adversarial network includes a generator (G network) and a discriminator (D network), the input of the G network is a low-resolution image, and the output is a super-resolution image, the input is processed through N residual blocks to obtain a reconstructed high-frequency image, and the output is obtained by superimposing the reconstructed high-frequency image and the input image. The D network is a discriminator, the input is a super-resolution image and a high-resolution image, and the input is processed through multiple layers of convolution to finally output a probability value for identifying the two. By introducing a self-attention (such as channel attention or spatial attention) mechanism, the feature extraction capability of the adversarial network can be enhanced, so that the adversarial network can better focus on the key information area of the image.

[0082] The generator of the super-resolution generative adversarial network adjusts its own parameters to minimize the difference between the generated image and the real high-resolution image (i.e. content loss), and the discriminator adjusts its own parameters to improve its recognition ability of real images and generated images (i.e. adversarial loss). The total loss function can be represented as:

[0083] loss GAN =loss G +loss D .

[0084] Wherein, loss GAN is the total loss of the super-resolution generative adversarial network, loss G is the loss of the generator, and loss D is the loss of the discriminator.

[0085] To focus attention on the region containing important features in the image, a self-attention mechanism is introduced in the residual block of the generator. In an exemplary embodiment, as shown in Figure 10 each residual block of the super-resolution generative adversarial network includes, in sequence, a convolution layer, a batch normalization layer, a Mish activation function, a convolution layer, a batch normalization layer, and a self-attention module. The Mish function has better performance in smoothness, stability, and generalization, and its calculation formula is as follows:

[0086] Mish(x) = x*tanh(ln(1+e x )).

[0087] where x is the input parameter of the Mish function, tanh() is the hyperbolic tangent function, ln() is the logarithmic function, and e is the natural constant.

[0088] The hyperparameters of the super-resolution generative adversarial network are set, including the optimizer, the learning rate, the number of samples, the number of iterations, the loss function, etc. During the training process, the training set images are used to train the super-resolution generative adversarial network, and the validation set images are used to verify whether the trained super-resolution generative adversarial network is overfitting and test the performance of the network. After training, the test set images are used to test the generalization ability of the optimal super-resolution generative adversarial network.

[0089] In this embodiment, the original sample images in the training set are input into the generator and enlarged by 8 times, and the resolution-optimized sample images in the training set are input into the discriminator as real images and generated false images for comparison. Using the PyTorch framework, a super-resolution generative adversarial network structure file model.py is written. As shown in Figure 10 the low-resolution input of the super-resolution generative adversarial network is processed through 8 residual blocks to obtain a reconstructed high-resolution output image.

[0090] In an exemplary embodiment, a certain high-resolution sample image is as shown in Figure 11 the low-resolution original sample image corresponding to this high-resolution sample image is interpolated and enlarged by the Bicubic algorithm to obtain a result image as shown in Figure 12 the above low-resolution original sample image is processed by the super-resolution generative adversarial network of the present application to obtain a result image as shown in Figure 13 By comparison, it can be seen that the image reconstructed by interpolation does not supplement new effective information for the resonance peak part, and compared with the high-resolution sample image in Figure 11 its resonance frequency point differs by about 0.79MHz, and there is still a problem that the resonance frequency point is not accurate enough. While Figure 13 the image detail information is significantly increased, the structural texture features are more reasonable and clear, and the visual effect is closer to Figure 11the high-resolution sample image in the middle, whose corresponding resonance frequency point is only 0.05 MHz different from Figure 11 the actual resonance frequency point of the high-resolution sample image in the middle, whose corresponding resonance frequency point is only 0.05 MHz different from

[0091] Based on the same inventive concept, the embodiments of the present application also provide a microwave blood glucose sensor data processing device. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more microwave blood glucose sensor data processing device embodiments provided below can refer to the limitations of the microwave blood glucose sensor data processing method in the above text, which will not be repeated here.

[0092] In an exemplary embodiment, as shown in Figure 14 a microwave blood glucose sensor data processing device is provided, comprising:

[0093] The sampling module 1401 is configured to acquire sampling data of the microwave blood glucose sensor. The image conversion module 1402 is configured to convert the sampling data into an original image. The resolution improvement module 1403 is configured to obtain a resolution-optimized image by using a resolution improvement model according to the original image. The data conversion module 1404 is configured to convert the resolution-optimized image into resolution-optimized data.

[0094] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and its internal structure diagram can be as shown in Figure 15 The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store sampling data of the microwave blood glucose sensor. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a microwave blood glucose sensor data processing method.

[0095] Those skilled in the art can understand, Figure 15The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0096] In an exemplary embodiment, a computer device is provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0097] In an exemplary embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0098] In an exemplary embodiment, a computer program product is provided, comprising a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0099] The beneficial effects of the present application are as follows:

[0100] (1) The present application converts the collected S 21 parameter values into image form, and uses image processing technology for more complex feature extraction and pattern recognition. Through the visual processing of data, the super-resolution generative adversarial network can extract more useful information from the image, thereby improving the accuracy of blood glucose detection.

[0101] (2) The present application utilizes the advantages of the super-resolution generative adversarial network, generates high-resolution spectral images through training, and can infer the missing high-frequency information using the model, thereby "compensating" for the resolution loss caused by insufficient sampling points to a certain extent.

[0102] (3) The super-resolution generative adversarial network built in the present application introduces a self-attention mechanism, which can more effectively capture long-distance dependencies in the input data and enhance the model's attention to important features. Through the self-attention mechanism, the super-resolution generative adversarial network can learn important features in the data during the training process. Even in the case of insufficient sampling points, the model can maintain sensitivity to resonant frequency points. This mechanism enables the model to more accurately reconstruct missing or low-quality signals, enabling the microwave blood glucose sensor to maintain high measurement accuracy in different use environments, thereby improving the resolution of the sampling data.

[0103] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0104] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments of each method. In the embodiments provided in the present application, any reference to memory, database or other medium can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc.

[0105] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0106] The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, not all possible combinations of the technical features in the above embodiments are described, but as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0107] The principles and implementations of the present application are described in the specific examples herein, and the above examples are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation and application range will be changed. In summary, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method of processing microwave blood glucose sensor data, the method comprising: The microwave blood glucose sensor data processing method comprises: Obtaining sampling data of a microwave blood glucose sensor; Converting the sampling data into an original image; According to the original image, a resolution optimization image is obtained by using a resolution enhancement model; the resolution enhancement model is a super-resolution generative adversarial network built in advance according to a training sample set; the training sample set comprises a plurality of original sample images and a resolution optimization sample image corresponding to each original sample image; Converting the resolution optimization image into resolution optimization data.

2. The microwave blood glucose sensor data processing method of claim 1, wherein, Converting the sampling data into an original image specifically comprises: According to the sampling data, a two-dimensional index matrix is obtained by using an index mapping method; The two-dimensional index matrix is subjected to data normalization processing to obtain a scaling matrix; the data range of the scaling matrix is between 0 and 255; According to the scaling matrix, a color mapping algorithm is used to obtain the original image.

3. The microwave blood glucose sensor data processing method of claim 1, wherein, The method for building the resolution enhancement model comprises: A plurality of blood glucose samples with different concentrations are prepared, and microwave signal original data corresponding to each blood glucose sample is obtained respectively; For each microwave signal original data, low sampling rate and low resolution data and high sampling rate and high resolution data of the microwave blood glucose sensor are obtained respectively; the low sampling rate and low resolution data is a sampling result with a sampling point number less than 100, and the high sampling rate and high resolution data is a sampling result with a sampling point number greater than 200; Each low sampling rate and low resolution data is converted into an original sample image, and each high sampling rate and high resolution data is converted into a resolution optimization sample image; Each original sample image is subjected to sample expansion to obtain an input image set, and each resolution optimization sample image is subjected to sample expansion to obtain an output image set; According to the input image set and the output image set, a super-resolution generative adversarial network is trained to obtain the resolution enhancement model.

4. The microwave blood glucose sensor data processing method of claim 3, wherein, Before obtaining, for each microwave signal original data, low sampling rate and low resolution data and high sampling rate and high resolution data of the microwave blood glucose sensor, the method for building the resolution enhancement model further comprises: The microwave signal original data is sequentially subjected to abnormal point removal and smoothing filter processing.

5. The microwave blood glucose sensor data processing method of claim 3, wherein, Each original sample image is subjected to sample expansion to obtain an input image set, and each resolution optimization sample image is subjected to sample expansion to obtain an output image set, specifically comprising: Each original sample image is subjected to image flipping, contrast adjustment, saturation adjustment and Gaussian noise addition processing to obtain the input image set; Each resolution optimization sample image is subjected to image flipping, contrast adjustment, saturation adjustment and Gaussian noise addition processing to obtain the output image set.

6. The microwave blood glucose sensor data processing method of claim 1, wherein, Each residual block of the super-resolution generative adversarial network comprises, in sequence, a convolution layer, a batch normalization layer, a Mish activation function, a convolution layer, a batch normalization layer and a self-attention module.

7. A microwave blood glucose sensor data processing apparatus for use in the microwave blood glucose sensor data processing method according to any one of claims 1 to 6, characterized by The microwave blood glucose sensor data processing device comprises: A sampling module for obtaining sampling data of a microwave blood glucose sensor; An image conversion module for converting the sampling data into an original image; The resolution enhancement module is configured to obtain a resolution-optimized image according to the original image by using a resolution enhancement model, wherein the resolution enhancement model is a super-resolution generative adversarial network that is built in advance according to a training sample set, and the training sample set includes a plurality of original sample images and resolution-optimized sample images corresponding to each original sample image. The data conversion module is configured to convert the resolution-optimized image into resolution-optimized data.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the microwave blood glucose sensor data processing method in any one of claims 1-6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the microwave blood glucose sensor data processing method in any one of claims 1-6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the microwave blood glucose sensor data processing method in any one of claims 1-6.

Images